Search

4D cone-beam computed tomography (CBCT) is an important imaging modality in image-guided radiation therapy to address the motion-induced artifacts caused by organ movements during the respiratory process. However, due to the extremely sparse projection data for each temporal phase, 4D CBCT reconstructions will suffer from severe streaking artifacts. Therefore, to tackle the streak artifacts and provide high-quality images, we proposed a framework termed Prior-Regularized Iterative Optimization Reconstruction (PRIOR) for 4D CBCT. The PRIOR framework combines the physics-based model and data-driven method simultaneously, with powerful feature extracting capacity, significantly promoting the image quality compared to single model-based or deep learning-based methods. Besides, we designed a specialized deep learning model named PRIOR-Net, which can effectively excavate the static information in the prior image reconstructed from the fully-sampled projections at the encoding stage to improve the reconstruction performance for individual phase-resolved images. Both the simulated and clinical 4D CBCT datasets were performed to evaluate the performance of the PRIOR-Net and the PRIOR framework. Compared with the advanced 4D CBCT reconstruction methods, the proposed methods achieve promising results quantitatively and qualitatively in streak artifact suppression, soft tissue restoration, and tiny detail preservation. [ABSTRACT FROM AUTHOR]

Limited-angle CT is a challenging problem in real applications. Incomplete projection data will lead to severe artifacts and distortions in reconstruction images. To tackle this problem, we propose a novel reconstruction framework termed Deep Iterative Optimization-based Residual-learning (DIOR) for limited-angle CT. Instead of directly deploying the regularization term on image space, the DIOR combines iterative optimization and deep learning based on the residual domain, significantly improving the convergence property and generalization ability. Specifically, the asymmetric convolutional modules are adopted to strengthen the feature extraction capacity in smooth regions for deep priors. Besides, in our DIOR method, the information contained in low-frequency and high-frequency components is also evaluated by perceptual loss to improve the performance in tissue preservation. Both simulated and clinical datasets are performed to validate the performance of DIOR. Compared with existing competitive algorithms, quantitative and qualitative results show that the proposed method brings a promising improvement in artifact removal, detail restoration and edge preservation. [ABSTRACT FROM AUTHOR]

Purpose: The purpose of this paper is to present an end‐to‐end deep convolutional neural network to improve the dual‐energy CT (DECT) material decomposition performance. Methods: In this study, we proposes a unified mutual‐domain (sinogram domain and CT domain) material decomposition network (DIRECT‐Net) for DECT imaging. By design, the DIRECT‐Net has immediate access to mutual‐domain data, and utilizes stacked convolution neural network layers for noise reduction and material decomposition. The training data are numerically generated following the fundamental DECT imaging physics. Numerical simulation of the XCAT digital phantom, experiments of a biological specimen, a calcium chloride phantom and an iodine solution phantom are carried out to evaluate the performance of DIRECT‐Net. Comparisons are performed with different DECT decomposition algorithms. Results: Results demonstrate that the proposed DIRECT‐Net can generate water and bone basis images with less artifacts compared to the other decomposition methods. Additionally, the quantification errors of the calcium chloride (75‐375 mg/cm3) and the iodine (2‐20 mg/cm3) are less than 4%. Conclusions: An end‐to‐end material decomposition network is proposed for quantitative DECT imaging. The qualitative and quantitative results demonstrate that this new DIRECT‐Net has promising benefits in improving the DECT image quality. [ABSTRACT FROM AUTHOR]

• A novel encoder-decoder transformer-based generative adversarial network designed to suppress sparse view CT image artifact. • In Transformer, we utilized the multi-Dconv head transposed attention module, enhancing its ability of features extraction. • To improve structure and detail recovery performance, we adopted the gated-Dconv feed-forward network in Transformer. • Within the GAN learning framework, we adopted a discriminator to enhance the ability of the generator. Sparse view CT images are often severely degraded by streak artifacts. Numerous studies have confirmed the remarkable progress made by deep learning (DL) in sparse view CT imaging scenarios. However, the mainstream CNN-based methods are inefficient when capturing feature information in large regions. In this paper, a transformer based generative adversarial network (SVT-GAN), which is designed to efficiently suppress artifacts in sparse view CT images, is proposed. We leverage the advantages of transformer networks and adversarial learning into a framework to improve the quality of sparse view CT image restoration results. The generator is primarily composed of an encoder-decoder structure that relies on the transformer model to learn multiscale local–global representations and leverage contextual information derived from distant artifacts. Moreover, in contrast with the standard transformer model, we utilize the multi-Dconv head-transposed attention (MDTA) module to enhance the ability of the proposed approach to extract both local and nonlocal information and produce impressive structure and detail restoration results. To suppress the transformation of artifact features, the gated-Dconv feedforward network (GDFN) is utilized. Within the GAN learning framework, we employ a simple nine-layer network as the discriminator to enhance the ability of the generator to suppress artifacts and retain features. Compared with the recently developed state-of-the-art methods, the proposed model significantly reduces serious noise artifacts while preserving details on the AAPM and Real CT datasets. Qualitative and quantitative assessments demonstrate the competitive performance of the SVT-GAN. [ABSTRACT FROM AUTHOR]

X-ray computed tomography (CT) is of great clinical significance in medical practice because it can provide anatomical information about the human body without invasion, while its radiation risk has continued to attract public concerns. Reducing the radiation dose may induce noise and artifacts to the reconstructed images, which will interfere with the judgments of radiologists. Previous studies have confirmed that deep learning (DL) is promising for improving low-dose CT imaging. However, almost all the DL-based methods suffer from subtle structure degeneration and blurring effect after aggressive denoising, which has become the general challenging issue. This paper develops the Comprehensive Learning Enabled Adversarial Reconstruction (CLEAR) method to tackle the above problems. CLEAR achieves subtle structure enhanced low-dose CT imaging through a progressive improvement strategy. First, the generator established on the comprehensive domain can extract more features than the one built on degraded CT images and directly map raw projections to high-quality CT images, which is significantly different from the routine GAN practice. Second, a multi-level loss is assigned to the generator to push all the network components to be updated towards high-quality reconstruction, preserving the consistency between generated images and gold-standard images. Finally, following the WGAN-GP modality, CLEAR can migrate the real statistical properties to the generated images to alleviate over-smoothing. Qualitative and quantitative analyses have demonstrated the competitive performance of CLEAR in terms of noise suppression, structural fidelity and visual perception improvement. [ABSTRACT FROM AUTHOR]

Low-dose computed tomography (LDCT) holds great potential to reduce radiation dose damage. However, LDCT degrades the signal-to-noise ratio (SNR) of projection and compromises the reconstruction image quality. This paper proposes a deep residual-constrained reconstruction (DRCR) framework via learned convolutional sparse coding (LCSC) for LDCT imaging. DRCR consists of multiple iteration blocks, and each iteration contains three modules: image update, residual correction and convolution representation modules. First, the image update module updates the reconstruction image with a DL-based prior constraint. Then, the residual correction module attempts to accurately estimate the residual features after mapping the projection errors into the image domain. Finally, the LCSC network is applied for feature map and filter updating to further constrain the image update process. To further improve the performance of DRCR, the perceptual losses of the image domain and projection domain are considered in the optimized model. The results obtained on two datasets show the competitive performance of the proposed framework, with a 0.5 dB PSNR margin, 0.007 SSIM margin, and 1.7 FID margin on the MAYO dataset and a 0.5 dB PSNR margin, 0.011 SSIM margin and 2.4 FID margin on the UIH dataset. [ABSTRACT FROM AUTHOR]

• A ultiscale reweighted convolutional coding neural network (MRCON-Net) for reducing artifact noise in low-dose CT imaging is proposed. • In this work, w extend traditional CSC to its reweighted convolutional learning form. Furthermore, we use dilated convolution to extract multiscale image features, which allows our single model to capture the correlations between features of different scales. • To automatically adjust the elements in the feature code to correct the obtained solution, channel attention is utilized to learn appropriate weights. Low-dose computed tomography (LDCT) has become increasingly important for alleviating X-ray radiation damage. However, reducing the administered radiation dose may lead to degraded CT images with amplified mottle noise and nonstationary streak artifacts. Previous studies have confirmed that deep learning (DL) is promising for improving LDCT imaging. However, most DL-based frameworks are built intuitively, lack interpretability, and suffer from image detail information loss, which has become a general challenging issue. A multiscale reweighted convolutional coding neural network (MRCON-Net) is developed to address the above problems. MRCON-Net is compact and more explainable than other networks. First, inspired by the learning-based reweighted iterative soft thresholding algorithm (ISTA), we extend traditional convolutional sparse coding (CSC) to its reweighted convolutional learning form. Second, we use dilated convolution to extract multiscale image features, allowing our single model to capture the correlations between features of different scales. Finally, to automatically adjust the elements in the feature code to correct the obtained solution, a channel attention (CA) mechanism is utilized to learn appropriate weights. The visual results obtained based on the American Association of Physicians in Medicine (AAPM) Challenge and United Image Healthcare (UIH) clinical datasets confirm that the proposed model significantly reduces serious artifact noise while retaining the desired structures. Quantitative results show that the average structural similarity index measurement (SSIM) and peak signal-to-noise ratio (PSNR) achieved on the AAPM Challenge dataset are 0.9491 and 40.66, respectively, and the SSIM and PSNR achieved on the UIH clinical dataset are 0.915 and 42.44, respectively; these are promising quantitative results. Compared with recent state-of-the-art methods, the proposed model achieves subtle structure-enhanced LDCT imaging. In addition, through ablation studies, the components of the proposed model are validated to achieve performance improvements. [ABSTRACT FROM AUTHOR]

• Proposed a method to estimate dual-energy CT images from traditional single energy CT image together with a single view projection at a different energy level. • Deep-learning algorithm is used to perform material-decomposition-like operations. • The approach can provide superior material-specific images with significantly reduced noise compared with standard DECT imaging. [Display omitted] Dual-energy computed tomography (DECT) is of great significance for clinical practice due to its huge potential to provide material-specific information. However, DECT scanners are usually more expensive than standard single-energy CT (SECT) scanners and thus are less accessible to undeveloped regions. In this paper, we show that the energy-domain correlation and anatomical consistency between standard DECT images can be harnessed by a deep learning model to provide high-performance DECT imaging from fully-sampled low-energy data together with single-view high-energy data. We demonstrate the feasibility of the approach with two independent cohorts (the first cohort including contrast-enhanced DECT scans of 5753 image slices from 22 patients and the second cohort including spectral CT scans without contrast injection of 2463 image slices from other 22 patients) and show its superior performance on DECT applications. The deep-learning-based approach could be useful to further significantly reduce the radiation dose of current premium DECT scanners and has the potential to simplify the hardware of DECT imaging systems and to enable DECT imaging using standard SECT scanners. [ABSTRACT FROM AUTHOR]